On-demand ride-pooling has been studied frequently in recent literature. However, real-world applications of large-scale ride-pooling systems are rare and the breakthrough of comprehensive systems is yet to come. In this study, we simulate a ride-pooling service based on real-world demand data from Hamburg, Germany and evaluate the impact of traffic uncertainty on the system performance using a state-of-the-art ride-pooling algorithm. We propose strategies to remedy customer constraint violations on wait and travel times due to wrongly estimated travel times. We show that uncertainty leads to a trade-off between violating customer constraints and rejecting requests after prior acceptance.